The arrival of data and analytics at football clubs has brought a lot of introspection and changes to the way that clubs operate, especially in the area of player recruitment.
We are often treated to behind-the-scenes snippets from sporting directors and analysts. However, coaches have been conspicuous by their absence when it comes to discussing data and its impact on tactics, training and player development. The perception is that coaches are “receivers” of information rather than being pivotal to the creation and development of these systems.
Analytics FC sat down with former Swansea City and Pafos coach, Cameron Toshack to bust some of these myths and explore how coaches can use data directly to support their work.
For me, coaching is about developing people and their talent. We seek out opportunities to make a positive difference wherever possible. Analytics in the modern game is key to driving this objective.
There is a plethora of data available to football clubs now. This has raised many questions for me as a coach. On the one hand, I have found it important to remind myself of the limits of data within a club coaching setting. I feel that I must make a conscious effort to always ask, “Why am I using or sharing this information?” and “How will it impact performance?”
But on the other hand, there are clearly great benefits from using data within a club environment. During my career, I have come to learn that data and analytics can act as a catalyst to the coaching process as well as driving player understanding.
As a coach, then, I believe in being data-informed rather than data-driven. The former allows an open mind to approach each situation and enables a place for subjectivity. In this piece, I am going to outline six ways that I have utilised data within my coaching career. I believe that each of these six ways has improved me as a coach.
Working to Objectives
One of the principal benefits to using data in coaching is that it allows you to set clear KPIs that are both objective and measurable.
In 2019, I got the opportunity to take on an international experience working as Head Coach of Pafos FC in the Cypriot Premier League, which currently sits 15th in UEFA’s league rankings. Following discussions with the club’s Sporting Director, it became clear our thinking was aligned in terms of on-field objectives. Fundamental to this was the shared recognition that the club needed to recruit young players whilst improving the players they already had.
The use of data to support our work was central to the project. We decided that we could only achieve our objectives if we had a way to measure and accelerate improvement.
This was a difficult task to achieve. At this point, Pafos were facing a relegation fight. Both myself and the sporting director spoke constantly about innovative ideas to action this. The short-term aim was to make an immediate impact on performance. The medium-term objective was to change the football style into an attacking, possession-based model. Collectively, we planned to use data to measure and support these changes.
We utilised a number of metrics but to measure the short-term impact, we decided upon a three-game rolling xG analysis to help us focus on the underlying performances rather than just the results. This would allow us to see if we were heading in the right direction in terms of performances.
My assistant coach, Gary Richards, and I were confident that our game model and the principles we were coaching would enable the players to deliver a more offensive strategy. As the chart below shows, we got the team ticking relatively quickly and it wasn’t long before we were aligned to just below Championship winning form (also raising the win percentage from 23% to 42%). We finished the season with the highest league finish in the club’s history:
The xG analysis was something we also communicated directly to the players and it became a key tool to support our message when speaking in both team meetings and individually. The same messages were being relayed to the ownership group by the Sporting Director and the CEO.
Developing a Game Model
Using data as a coach can also help you settle on and monitor a game model that is suitable to the needs of your team.
In the interview process for the job, I was clear with the board about the game model and identity I wanted the team to have on the pitch, and how we would deliver that. Of course, having worked at Swansea City for nearly 7 years there are some heavy influences from my time there. That approach seemed to chime with the direction the club wanted to go in. However, in discussions with the sporting director—and along with input from coaches and performance staff—we knew that it would be important to have some way of measuring that.
The first job was to develop KPIs to measure the overarching playing style of the team. We were lucky enough to have access to Analytics FC data and performance algorithms which could help us map where we were and how we were developing. We tried to keep the concepts simple to appeal to the staff, players and owners.
The next layer of analysis we used was to look at tendencies or “principles of play” and also to look at how our principles were impacting the opposition’s playing style. This gave us some immediate and objective insight into which areas of our game model were on track and which needed to be worked on.
In discussion with Analytics FC, we also created some bespoke and useful internal tools which could be deployed by our performance and video analysts in training and matches. Here we focused on three key zones:
Zone A is the area between the opposition’s defence and midfield. This aspect of the game model was about breaking the lines and working attacks through the central areas of the pitch:
The second zone—Zone B—is the wide areas either side of the defensive block. Our game model focused on fashioning 1v1s and then going direct to goal to generate dangerous moments in the penalty area:
The final zone—Zone C—was the space in behind the defence. We looked to get players in behind the defensive line to generate chances with the opposition defenders running backwards.
With these three zones targeted, data analysis presented us with a way of assessing how successful we were being in each area. For example, we could quickly and easily determine where our 1v1s were taking place, how many of them were successful and how many were resulting in dangerous chances.
In this way, data and coaching went hand in hand and made my job as a head coach much more manageable.
Finding players to fit the style
Identifying the player characteristics needed to deliver a style of play is a key skill for any head coach. Data analysis can streamline this process, presenting coaches with a statistical profile which could indicate a broader skillset than was initially considered.
The transition of Jason Puncheon to a deep-lying playmaker at Pafos was exactly this type of scenario. Puncheon’s attributes were a great fit to deliver the style of play that was key to the game model. After a difficult start to the season, he became a key contributor to the team’s success. By the time the season came to a close, he was being spoken about as a potential Team of the Season candidate in the role he transitioned into.
Of course, there is hard work that goes on in the background to make a positional transition like this, not least from the player himself, who needs to buy into the process. Daily coaching and feedback from staff are also essential.
We were supported in this process by Analytics FC and their TransferLab data model. Below you can see a before and after analysis which was used as a monitoring tool to assess player performance and development over time:
Our game model required a player who could dominate and dictate from a deep position. This is where we focused our training and match tactical work together with Jason. His vast improvements in short (progressive) passes and line-breaking passes were both key traits required for a deep playmaker in our game model.
Here are some best practice examples of the role Jason was asked to play:
Notice his ability to drift backwards into space, before carrying the ball and breaking the line with his passing. This became a regular sight for us that season.
Building player profiles to aid performance and recruitment
At Pafos, these sorts of data profiles of our players soon became the norm, and with the sporting director, the other coaches and analysts, we would eventually build in-depth player positional profiles for all the positions across our game model.
In order to get the best out of the data, we matched our positional traits with Analytics FC’s metrics as much as possible to allow us to track and analyse current performance over a set number of games and compare to the required level set within the club’s objective framework.
We headlined game moments in possession (build up/creating the attack/finishing the attack), the two transitional moments (move to attack/move to defend) and the out-of-possession moments (high press/medium press/box defence). Within each section, the traits were then measured via the data profile of each player.
These measurements were benchmarked at both league level and against similar level leagues across Europe. This had the dual purpose of measuring internal player performance but also offering a way of benchmarking against recruitment targets.
Individual player development
Using the positional data, we could then produce reports to aid our discussions with individual players. Together with my multi-disciplinary team (MDT)—assistant coaches, the head of performance, the head of analysis, the lead physio and the fitness coach—we carried out regular individual player reviews.
As the season progressed, the reports were updated by the MDT. A regular meeting between all MDT staff was also carried out to assess the progress of players and discuss any issues or ideas which may have arisen. Individual Development Plans (IDPs) were then devised for each player.
These IDPs formed the basis of an interactive conversation between coach and player who discussed the content through verbal, visual (video) and visual (data). The relevant members of staff then took responsibility for supporting the player around specific individual interventions that were required to help the player achieve his targets on the positional profile.
Creating individual targets to monitor progress
Once again, data analysis offered us a useful benchmark against which to assess our players through the season, giving us an objective to measure performance and recognise improvement.
Jerson Cabral (former Feyenoord and FC Twente) and Vladmir Etson “Va” (an Angolan International) offer great examples for how this worked in practice. In possession, we wanted to drive the team towards a more aggressive and attacking brand of football. The patterns of play that Gary and I instill in our wide players—which have proved successful in the past with players such as Daniel James at Swansea—were perfectly suited to helping improve these two players.
Below you can see two examples of this type of focused pattern of play:
Starting with a switch pass from the defensive line, the relationship between the full back and the wide player is explored, learned, and then, the expectation is that these moments get transferred onto the pitch to aid players’ decision-making in key moments.
The analysis of Carbal and Va’s improvements was very pleasing to see. The analysis below certainly reflected the feelings of the coaching staff and the players.
Data is vital to supporting the development process, it helps players in learning and understanding their roles and also adds a layer of accountability (for both the player and the coaches).
In addition, numbers have few cultural barriers. Having previously coached at Wydad Casablanca in the African Champions League and the North Macedonian National Team, I’d seen first hand that you need strategies to effectively translate your coaching ideas to players. The Pafos squad contained players speaking seven different languages. I felt that objective data facilitated conversations that might otherwise not have happened, again improving the players and the team.
Our biases as coaches need to be assessed regularly. The use of analytics and data to balance our opinions about football is a helpful way of doing this.
Often as coaches, we hear supporters referring to certain out of favour players in glowing terms. Usually, it’s the positives that are recalled in these moments. More often than not, though, when the numbers are crunched, a very different picture appears.
Using data can help combat external pressure to mix things up on the field. At Pafos, the data was a very important aspect of our approach to running a club at a point in time where a wrong step could have resulted in relegation. Instead, we trusted the process and avoided giving in to short-termism or opinion-heavy assessments of what we should do.
Data analysis is also important for organisations during periods of contract renewal. Data can allow us to question our thinking as coaches, reflect and revisit with an open mind to reduce such bias.
I opened this piece by saying I believe in being data-informed rather than purely data-driven. The former allows more of an open-minded approach to each situation and enables subjectivity to have a place.
By combining a high level of emotional intelligence and selecting the appropriate data you can certainly drive progress. But it doesn’t always take the form of analysing numbers or reviewing video. Sometimes, it comes in the form of analysing a player’s mood, his behaviours.
When it comes to football, I believe it’s about the person first and then the player. But on-pitch statistical data has become a very real and very helpful tool for coaches like myself to get the best from people in the football context. As the saying goes “if it gets measured, it gets done”.